perceptual modalities
TRANSCRIPT
Marek McGann
Perceptual ModalitiesModes of Presentation or Modes of Interaction?
Abstract: Perceptual modalities have been traditionally considered
the product of dedicated biological systems producing information for
higher cognitive processing. Psychological and neuropsychological
evidence is offered which undermines this point of view and an alter-
native account of modality from the enactive approach to understand-
ing cognition is suggested. Under this view, a perceptual modality is a
stable form of perception which is structured not just by the biological
sensitivities of the agent, but by their goals and the set of skills or
expertise which they are deploying at a given time. Such a view sug-
gests that there is no such thing as an experience that is purely visual,
auditory, or otherwise modal and that our attempts to understand
consciousness and the mind must be conducted within a framework
that provides an account of embodied, goal-directed adaptive coping
with the world.
Keywords: Modality, perception, sensation, enaction, skill theory,
consciousness
Introduction
This paper is about perceptual modalities — how we should conceive
of them and what their relationship is to our bodies and to our
consciousnesses. Addressing such a topic, however, first demands
that we get straight a common distinction between ‘sensation’ on the
one hand, and ‘perception’ on the other.
Cognitive scientists tend to interpret the distinction between sensa-
tion and perception in one of two ways. The first makes a strong
Journal of Consciousness Studies, 17, No. 1–2, 2010, pp. 72–94
Correspondence:Marek McGann, Department of Psychology, Mary Immaculate College, Univer-sity of Limerick. Email: [email protected]
distinction between the two, making the former a physiological pro-
cess and the latter a psychological one. In this case, sensation is a
characteristic of various surfaces and cells of the body which react in
particular ways when they make contact with particular forms of
energy such as light, pressure, or sound, or with certain chemicals.
Perception, on this strong view, is the psychological process, enabled
by sensation, by which we come to some form of direct knowledge of
the world.
The second form of the sensation-perception distinction found
amongst cognitive scientists makes the division between the two less
clear. In this case, sensation is the bare basis of later perception, the
recognition of elementary aspects of a given object or stimulus such as
its form, colour or pitch. Perception, then, is the process by which
these elements are assembled or bound into coherent but complex
wholes, transforming the sensations of redness and roundness into the
perception of a ripe tomato, for instance. This second form of the dis-
tinction, though both popular and quite intuitive, begs for confusion.
The physiological aspects of sensation are less clearly articulated,
though generally remaining at the foundation of such basic aspects of
perception as recognition of colour and form. In this second case,
then, the concept of sensation straddles the division between physiol-
ogy and psychology, and thus, I suggest, obscures some very impor-
tant aspects of that division.
The focus of this paper will be on perceptual modalities, and their
relationship to sensation in that first, clear sense of physiological
response. The differences between modalities form some of our most
basic understandings of the structure of consciousness, and indeed the
very nature of consciousness. The traditional view of the nature of
perceptual modalities sees not only a strong continuity between the
physiological response to a stimulus and the resultant perception of it,
but a near identity between the two — an inextricable relationship
illustrated by the popularity of the second, more vague form of the
sensation-perception distinction. In this paper I will draw on both psy-
chological and neuroscientific research to undermine some of our
most fundamental intuitions concerning the relationship between sen-
sation and perception and thus to undermine the long-standing tradi-
tional account of perceptual modalities, an account which can trace its
lineage directly to Johannes Müller’s doctrine of ‘specific nerve ener-
gies’, put forward in 1826. In the place of this traditional account I
will argue for an enactive account of perceptual modality, one which
draws heavily on the work of Kevin O’Regan, Alva Noë and Erik
Myin amongst others (Myin & O’Regan, 2002; Noë, 2004; O’Regan
ENACTING MODALITIES 73
&; Noë, 2001a) but which offers a more coherent and more complete
story of the various elements of a modality.
The enactive approach to cognitive science has been knocking
around for a while now, but has thus far been largely relegated to the
domain of ‘low level’ cognition, the basic aspects of perception and
immediate bodily action. The account of modality that I will advance
later in this paper will provide grounds to reject a clear distinction
between lower and higher cognition, and attempt to show how an
enactive approach can equally and fruitfully be applied to our under-
standing of the richer and more complex forms of cognition.
The term ‘enactive’ has become something of a buzz word in the
recent Cognitive Science literature, and as such has developed a few
different meanings. In the present paper, I am applying the term in the
specific sense proposed by Varela, Thompson, & Rosch (1991), and
developed in subsequent works by several authors including Varela
(1997), Di Paolo (2005), Thompson (2007), Di Paolo et al. (in press)
and others. This approach is in many ways a continuing development
of the phenomenology of Merleau-Ponty, and the integration of some
of his insights with more recent developments in the dynamics of biol-
ogy and autonomous systems (see Varela, 1979; Weber & Varela,
2002; Thompson, 2007).
The emphasis on biological autonomy and the dynamic interaction
between such an autonomous system and its environment in the expla-
nation of cognition means that the enactive approach argues for a con-
tinuity between the basic processes that underlie life, adaptive
response to the world, and cognition. The mind is seen not as compris-
ing a set of distinct processes driving perception, cognition and
action, but a complex of skills which allow an autonomous agent to
maintain itself and achieve its goals in interaction with its environ-
ment. Perception, cognition and action are three facets of this single
process of adaptively coping with the world (what the enactive litera-
ture refers to as ‘sense-making’) rather than being distinguishable
links in a chain of processes that begin with ‘input’ at the sensory sur-
faces and end with ‘output’ at the muscles. Though the mind is driven
by the autonomy of the agent, all of its activities occur in engagement
with those aspects of the world around it that impinge upon it. The
description of either an agent or its environment will therefore always
perforce involve reference to the other – the cognitive agent and its
environment are inextricably entwined and an analysis must appreci-
ate not just one element or the other, but how the two interact and are
interrelated.
74 M. McGANN
Such a perspective makes the approach closely related to what I will
refer to as ‘dynamic sensorimotor’ accounts such as those of Susan
Hurley (1998; see also Hurley & Noë, 2003) and the theory of percep-
tion developed by O’Regan & Noë (2001a) and Noë (2004). Much of
this paper is a development of the implications of that theory. But the
enactive approach’s emphasis on the autonomy of the acting agent and
the emergence of cognition in the interaction between that agent’s
autonomous values and its environment, is not clearly laid out in a
dynamic sensorimotor view. Though the enactive approach used here
would involve a dynamic sensorimotor perspective, it is not clear that
the dynamic sensorimotor theories would adopt all aspects of an
enactive view.
The approach used here has a somewhat converse relationship with
the work of Natika Newton (1996), who strongly emphasises the
goals and values of cognitive agents and their essential (and often
overlooked) role in accounts of cognition. Newton, however, also
describes cognition as occurring in the embodied, sensorimotor repre-
sentations in the brain, whereas the enactive approach used in this
paper locates cognition only in the actual interaction between the
agent and its environment (for a more thorough exploration of the
‘where’of enaction, see Di Paolo, 2009). Such an enactive view, as we
shall see, provides some dramatically counter-intuitive implications
for even basic concepts of mind such as perceptual modality.
Perceptual Modalities
In traditional, and intuitive, ways of thinking, a perceptual modality is
a ‘mode of presentation’ of a particular stimulus. We might encounter
an object in many different ways, but one basic and simple aspect of
any perception will be the mode — visual, auditory, haptic etc. — in
which it is presented to us as perceivers.
The theory of modalities currently dominant in the cognitive sci-
ences is effectively that originally put forward by Johannes Müller on
‘specific nerve energies’. The essential idea is that sensory neurons
are responsive to particular forms of energy and it is this specificity in
neuronal response that gives a modality its character. The cells of the
retinae, for example, are specifically sensitive to light, and not, for
example, to temperature or sound, while the neurons of the cochlea
and ear are sensitive to the pressure waves of sound, but not to light.
This is a physiological explanation of perceptual modality that fits
well with our intuitions of clear distinct and basic modes of percep-
tion. Each modality, in this view, will have its own dedicated
ENACTING MODALITIES 75
anatomical organs — sight has the eyes, hearing the ears, taste the
tongue and so forth.
Modern neuroscience uses a version of this concept that is only
slightly modified, and is seen to have been ‘conclusively demon-
strated’ (Kandel et al., 1995, p. 371). This slight modification extends
the organs of interest from the sensory surfaces deep into the brain,
where we can find more anatomy particularly sensitive to and appar-
ently dedicated to dealing with, stimulation of a specific kind. These
neural organs appear to take sensory information in its raw form at the
sensory surfaces and pass it up to more complicated cross-modal pro-
cessing streams which are the medium of central, multi-modal cogni-
tion. It is processing in these more narrowly dedicated systems that is
believed to give a perception its specific modal character.
This standard view of modalities has more than a century and a half
of support and is pretty much unquestioned in current mainstream
Cognitive Science.
Just how many such modalities there are is not quite so clear,
though, even in the mainstream. Whatever the truth of the matter, it
would appear that we have more than the five traditional modalities.
Proprioception, the vestibular sense and kinaesthesia, for example,
extend beyond the typical conceptions of touch, and our perceptions
of texture, temperature and pain are all apparently underpinned by
separate neural systems. But this debate on the number of modalities
has not undermined the strong continuity between sensation and per-
ception inherent in the traditional view. In fact, the idea that the senses
can be distinguished and counted depends on the traditional model,
which make modalities modular — separated by separate organs and
neural bases, without cross-talk at least at the low levels of early
perceptual processing.
Research in both psychology and neuroscience, however, would
lead us to re-evaluate these basic elements of the traditional account of
modalities. In doing so, we find ourselves facing some important and
intriguing questions about the structure and form of consciousness.
Questioning Modality Modularity
Several cross-modality illusions indicate the presence of inter-modal-
ity influence. This begins to beg questions of the traditional view: if
the nerve energies are that specific, then there should be significant
insulation between modalities at least at the lowest levels. Modality
modularity should be a given. Probably the most commonly known
cross-modal illusion is the McGurk effect, first reported by McGurk
76 M. McGANN
& McDonald (1976). The McGurk effect is a phonological
misrecognition based on conflicting auditory and visual information.
Quite simply, while hearing one phoneme (such as ‘ba’) repeatedly
and watching someone mouth another (such as ‘ga’) at the same rate,
we can misperceive the sound as an intermediate phoneme (such as
‘da’). There are a number of further findings about the McGurk effect,
including that it occurs not only in adults but in children and that some
sounds are better than others for producing it. Because it is associated
with the articulation of phonemes, however, it is not the best evidence
for truly low level interaction between perceptual modalities. After
all, language is at least implied here, and thus the interaction is likely
to be the result of learning the speech sounds of our native tongues.
Such learned phenomena are usually taken to be higher level psycho-
logical phenomena, and will convince very few that we need to doubt
the kind of modularity of modalities that the specific nerve energies
theory tends to imply (though the McGurk effect has also been shown
in pre-linguistic children; Rosenblum et al., 1997). There are other
inter-modality interactions, however, some of which are more starkly
problematical for a traditional view. Shams et al. (2000; 2001) have
reported a cross-modal illusion they refer to as the ‘illusory flash
effect’. In the case of the illusory flash effect, a single brief flicker of a
visual stimulus is presented simultaneously (within 100ms) with two
auditory beeps. The result is the perception of two flickers of the
visual stimulus. Like many illusions, this effect is robust, resisting
practice and prior knowledge on the part of the perceiver. Violentyev
et al. (2005) describe a similar illusion induced by tactile stimuli
rather than auditory ones.
Shimojo & Shams (2001) review this phenomenon amongst a range
of others and conclude that the traditional modular view of sensory
modalities doesn’t hold up to proper inspection. Their conclusive title,
‘Sensory modalities are not separate modalities’, seems a bit strong to
our intuitions though — while we perceive the world as a rich weave
of sensations, visual, auditory, tactile and more, we do not perceive it
as a smear of indistinguishable modes of sensation. Such inter-modal
influence, however low level, may still not convince us that our expe-
rience of perceptual modalities is not simply dependent on the differ-
ent ways in which sensory information is transformed by our nervous
system as it is processed towards some multi-modal integration in the
brain. Striking work by the neuroscientist Walter Freeman, though,
makes a deeper and more profound challenge to a simplistic specific
nerve energies approach to modality. If Freeman is correct, then at the
very early stages of perceptual processing, the details of sensory
ENACTING MODALITIES 77
stimulation are lost, disrupting another aspect of the traditional view,
our intuitions about the continuity between sensation and perception.
Questioning Specific Physiological Bases
Neural dynamics, sensation and perception
Because in the traditional view the particular qualities of a modality
somehow arise from the specificity of nervous sensitivity, we should
expect unique neural pathways to maintain that specificity of sensitiv-
ity into the brain — the modern extension of Müller’s specific nerve
energies doctrine. The modalities of perception are therefore, under
the traditional view, entirely dependent on the character of sensation.
However, the work of Walter Freeman and his colleagues on the neu-
ral dynamics of perception cast some doubt on this possibility. In par-
ticular, Freeman has shown a Rubicon in perceptual processing, a
point where in the modalities of smell, touch, sight and hearing, the
sensory signal is completely lost, and perception determined not by
the pattern of stimulation at the sensory surface, but by its perturba-
tion of the inherent activity of the brain itself — it’s not just what the
sensory organs are doing, but what the brain is already doing, that is
involved in perception. Freeman’s work was conducted on rabbits,
and began with investigations into the neurophysiology of smell
(Freeman, 1991; 2000). Using internal measurement of the olfactory
bulb, Freeman showed that the activation of the bulb for similar stim-
uli in different reinforcement contexts was remarkably different. That
is, similar sensations did not necessarily produce similar activations
of the olfactory bulb. What is more, the patterns of activation for some
stimuli would change when other stimuli were given value through
reinforcement.
Figure 1. shows patterns derived from EEG readings of the bulb for
a rabbit before and after a new reinforcement contingency was intro-
duced the for the scent of amyl. Note particularly that the pattern asso-
ciated with air changes considerably after the change in reinforcement
condition despite the fact that no change in the value of the smell of air
was involved (the smell of air, in this case, is simply the background
smell of the rabbit’s cage). Freeman uses these and related findings to
argue that the patterns of response of the bulb to a given stimulus have
a lot more to do with the state of the bulb than they do the actual sen-
sory activity evoked by the stimulus. Because of the rich feedback
dynamics within the neurons of the olfactory bulb the activity on the
sensory surfaces cannot strongly determine the activity of the bulb.
Rather, sensory activity perturbs the continuously ongoing activity of
78 M. McGANN
the neurons. The patterns of activity in the bulb are largely the product
of its own activity and history. All sensation does is give it a nudge
across its own landscape of attractors. That perturbative nudge must
occur in a context of both recent activity of the bulb and the fact that
the landscape of its dynamical phase space is constantly changing.
The important point here is that what appears to have determined what
was smelled by the rabbits in Freeman’s studies was the rabbit’s own
history and the history of various reinforcement contingencies. The
sensory information per se plays a much attenuated role in forming
higher level neural patterns. What is processed by the perceptual
cortex is categorical, the meaning of the stimulus for the individual,
not the raw stimulus itself (Freeman, 1991). Perception does not
appear to be strongly determined by the specifics of the sensory level.
ENACTING MODALITIES 79
Figure 1. Surface EEG patterns for olfactory bulb responses to air and amyl
nitrate under different reinforcement conditions (adapted with permission
from Freeman & Schneider, 1982, ‘Changes in spatial patterns of rabbit
olfactory EEG with condiitoning to odors’, Psychophysiology, 19(1), pp.
44–56). Note that no change in reinforcement to air was involved, rather the
reinforcement of amyl alters the patterns of activation associated with air.
Freeman & Barrie (1994) generalised this point to the neural pro-
cessing of touch, audition and vision, once again in rabbits. What this
work shows is a Rubicon between sensation and perception which
means that wherever smell gets its odour, it is not from the activity on
the sensory surfaces of the nose. We may speak of ‘sensory modali-
ties’ but these are not the same modalities in which we perceive. This
is a profoundly counter-intuitive point, but it is one which we will see
is implied by the enactive approach to perception and which I am sug-
gesting is implied by the work of Walter Freeman and his colleagues.
The neurodynamics of sensation and perception undermine our basic
beliefs concerning the modalities of perception. The weak conception
of the sensation-perception distinction is challenged by these find-
ings, and deep problems open up for any account of perception which
relies heavily on an implicitly held continuity between the two. Note
that suggesting that the neural organs specific to different forms of
sensory stimulation are widely distributed, rather than anatomically
localised in the brain does nothing to attenuate the implications here.
What matters to the specific nerve energies view is that the neural
pathways are dedicated to the processing of the sensory information,
but Freeman’s studies show that that sensory information is irrevoca-
bly lost once it comes in contact with the complex, multiply connected
and autonomous activity that is already occuring in the brain.
What is more, that activity is not insulated from the broader interac-
tion between the animal and its environment. Nothing changes in the
amyl nitrate chemical itself during this research — it still interacts in
the same way with the olfactory receptors. But the implications of that
interaction, its value to the animal and its consequences for what the
animal is doing transforms the way the olfactory bulb responds not
just to that specific sensory signal, but to others too (as illustrated, the
background smell of the rabbit’s cage appears affected, despite not
being involved in a change in reinforcement value). This suggests that
what is going on is not some ‘top-down’ hypothesis of prediction
about what the signal might be, but a coordination of the rabbit’s
activity with the sensation. Perception is a complex of motivations
and on-going activities of the cognitive agent that is contextualised by
the animal’s needs, its behaviour and its history – it is not the output of
a progressive chain of processing, an assembly-line of sensations. Nor
is it a neutral prediction of what the sensation might be, given that the
nervous system’s responses alter not to changes in smells, but to
changes in their implication for the animal.
All this said, nor is perception arbitrary or unrelated to sensation.
Perception appears to involve coordinating with the world, not
80 M. McGANN
imposing any old interpretation upon it. We don’t just get to smell
what we want to, or even what we expect to. The question arises then,
if the character of perception is not determined by the sensory organs
being stimulated, what does determine it?
Freeman’s research suggests that somehow, it is not the eyes that
matter for vision, nor the nose for smell, but the right kinds of interac-
tion between an acting perceiver and the world. This is not to say that
normal vision does not depend on the eyes — that much is obvious —
but is to suggest that that even in the normal case vision is not just a
matter of what happens in the movement of neural activation upstream
from the retinae, and it is not necessarily dependent on optical input at
all. Evidence in support of such thinking, which I believe leaves the
traditional account of perceptual modalities wide open for assault by
the enactive approach, is the much quoted matter of sensory
substitution.
Sensory substitution
In work much cited within the enactive literature (e.g. (e.g . O’Regan
& Noë, 2001a,b; Noë, 2004; Hurley & Noë, 2003), neuropsychologist
Paul Bach-y-Rita (1972, see also 1984) has produced vision-like
experiences in both blind and normal participants, based entirely on
sensation on the skin. This remarkable effect is achieved using a
system referred to as TVSS (tactile-visual-substitution-system). An
array of vibrating contacts are placed on the skin (on the belly, back or
tongue) of the person. The activity of the array is controlled by a
camera which the person wears on their head. Bach-y-Rita (1972)
describes how, given a period of time actively exploring their environ-
ment with this system, participants (blind or blind-folded sighted peo-
ple) have reported experiencing depth, occlusion, basic object
recognition, looming and even one form of visual illusion (the water-
fall illusion, Bach-y-Rita, 1984). It is crucial that in the learning
phase, the participant have control over the camera — the sensation
alone is not enough. The person wearing the camera and vibrator
array must have the opportunity to integrate their tactile experiences
into a deliberate and on-going interaction with their environment.
Due to the limitations of skin sensitivity, the resolution of the vibra-
tor array is never very high (about 20x20; see Bach-y-Rita, 1983, for
discussion) and as a result the ‘vision’ it provides is weak indeed.
Nevertheless, the experience involved does appear to be visual (or at
least distal and spatial) in character, and is certainly not experienced
as tactile by the individuals in question. This despite the fact that they
ENACTING MODALITIES 81
can, if they wish, pay attention to the vibrations on their skin. They
generally don’t pay much attention after the training period, though,
and these sensations seem to be relegated to the same level of percep-
tion we normally have for the clothing or watch that we might be
wearing.
Neural dynamics work by Freeman and his colleagues, along with
such phenomena as TVSS, constitute a significant challenge to our
traditional conceptions of perceptual modalities. The question then
arises of how we might cope with this challenge. If the specific nerve
energies account of perceptual modalities is wrong, what kind of
account might we use to replace it? An enactive approach to percep-
tion, which would endorse such work as that of O’Regan & Noë
(2001a), Noë (2004), O’Regan et al. (2005) offers us a new means of
conceiving the differences between the modalities of perception.
The Enactive Approach to Perceptual Modalities
The basic conception of perception within the enactive approach is as
a sensitivity to the aspects of the world that have meaning for the
agent’s actions. At first blush, that seems like either a very tall order or
a rather vague promissory note. This general idea, however, makes
clear the fundamental and irreducible circularity of the relationship
between perception and action. We perceive, not as a broad hoovering
up of available information from the world, but in order to act. What
we perceive is determined as much by what we are trying to achieve as
what our sensory systems are capable of. We don’t just wait for the
world to come to us, nor do we passively accept information available
at the sensory surfaces.
A complete account of perception within the enactive approach is
beyond the scope of the present paper. We can briefly introduce our-
selves, though, to the aspects of such an approach that have a bearing
on our conception of modalities.
The dynamic sensorimotor theory of vision put forward by
O’Regan & Noë (2001a; see also Noë. 2004) is a strong foundation on
which to build a thoroughly enactive approach to perception. They
claim that to perceive is to exercise a mastery of the sensorimotor con-
tingencies of a given situation. Sensorimotor contingencies (SMCs)
are regularities in the interactions between our bodily movements and
our sensations. A very simple example is if you move your eyes to the
left, the pattern of stimulation on the retinae moves to the right.
Another is if you move toward an object, then the retinal stimulation
from points of texture on the object move outwards from the centre
82 M. McGANN
toward the periphery of the retinae. The important point here is that
the regularities that constitute these sensorimotor contingencies are
not simply part of the world, nor part of our sense organs, but arise
because both the world and our sense organs are relatively stable, and
therefore the interactions between the world and our sense organs will
have relatively stable characteristics. When we exercise a mastery of
these SMCs, when we can confidently and reliably guide our bodily
movements appropriately to perform actions in a given environment,
we are perceiving.1
In discussing sensorimotor contingencies, O’Regan & Noë (2001a)
distinguish between SMCs based on the object and those based on the
sensory apparatus. An example of an object-based visual SMC would
be that as a viewpoint moves around a mug, the handle appears, dis-
torts and disappears in characteristic ways. An example given of an
apparatus-based visual SMC is that as the eyeball moves a straight
line produces different forms of curves of stimulation on the retina
because of the shape of the eye and retina. O’Regan & Noë (2001a,
p. 946) identify apparatus-based SMCs as a significant original con-
tribution, something not present in previous action-focused accounts
of perception such as McKay’s and Gibson’s. The concept offers a
strong reminder of the importance of embodiment and bodily interac-
tion with the world.
Finally, O’Regan & Noë (2001a, p. 943) suggest that the difference
between sensation and perception is grounded in the difference
between object- and apparatus-based sensorimotor contingencies.
While it would appear (from evidence such as TVSS) that they are
correct in throwing into relief the role and the details of the physical
body for perception, the object-based versus apparatus-based SMCs
distinction doesn’t quite hold up to scrutiny. Take their example of the
distortion of stimulation on the retina given movements of the eyeball
— as the eyeball moves upward the stimulation on the retina from a
straight line in front of the person transforms from a straight line
(when the eye is focused directly on the line) to a curved one (when
the line is below the point of focus) because of the shape of the
ENACTING MODALITIES 83
[1] In more recent work, Noë (2004) prefers the phrase ‘knowledge of sensorimotor regulari-ties’. Reference to knowledge along with mastery is also made in the original O’Regan &Noë (2001a) paper. Precisely what kind of knowledge is involved, however, remainssomewhat unclear, and has been a source of criticism for the approach (e.g. Hutto, 2005;Rowlands, 2005). I have opted for the more skill-focused concept of ‘mastery’ as I find itthe more compelling account. The implications concerning perceptual modalities devel-oped later in the paper do not depend completely on this choice, however, and can also bedrawn out from a more knowledge-focused description of this dynamic sensorimotoraccount of perception.
eyeball. Those distortions of stimulation cannot be reliably contingent
without at least implicit reference to objects, or some external thing.
In the example given, the reference is an explicit one, to a straight line.
And there will always have to be such reference: while the idea of the
distinctiveness of apparatus-based versus object-based is sensori-
motor contingencies is a useful one to make clear the role of the details
of embodiment in perception, it should not be asked to do any real the-
oretical work. There is a distinction between sensation and percep-
tion, but it is not captured by the distinction between object- and
apparatus based SMCs. This does not cause immediate problems,
however, because O’Regan & Noë (2001a) do not really ask their def-
inition to do important theoretical work.
The sensorimotor aspects of the relationship between the agent and
their environment cannot be the whole story. That sensorimotor con-
tingencies enable perception is certainly consistent with the enactive
approach, but it lacks any reference to the endogenously driven activi-
ties of the agent — the valued, goal-oriented nature of an agent’s
actions. Perceiving is not sensorimotor contingencies alone, but the
exercise of their mastery. It is for this reason that O’Regan & Noë’s
view is considered a form of ‘skill theory’ of vision.
So how are modalities constituted within such an account?
O’Regan & Noë (2001a) claim that SMCs are governed by relatively
stable sets of laws, and that it is these more or less coherent sets of
laws that constitute a perceptual modality.2 As with the contingencies
themselves, the laws governing contingencies emerge because of the
details of our specific embodiment and the kinds of actions which our
environment enables. Some movements will affect the pattern of sen-
sation on the retinae, but have no effect on other sensations, e.g. a
movement of the head, which produces dramatic changes in retinal
stimulation but will have no effect on the feeling of a wine bottle in
our hands. Other movements will affect our sensations of touch (such
as moving our hands over that wine bottle), but have no effect on
vision. Below is a table, taken from an on-line draft of O’Regan et al.
(2005) which gives some examples of sensorimotor contingencies for
vision and hearing, and how these contingencies can offer an explana-
tion of the differences between the two modalities.
84 M. McGANN
[2] O’Regan & Noë (2001a) actually refer to the laws governing SMCs as constituting sen-sory modalities, but draw no explicit distinction between sensory and perceptual modali-ties. It seems clear, however, that if sensation is the transduction of specific kinds of ener-gies into neural activity then sensory modalities will be physiological and organ-specific –much closer to Müller’s view. O’Regan & Noë appear to acknowledge this in pointing outthat there are two different kinds of contingencies, apparatus-based contingencies andobject-based contingencies, as mentioned.
Action Seeing Hearing
Blink Big change No change
Moves eyes Translating flowfield No change
Turn head Some changes in flow Left/right ear phase and
amplitude difference
Move forward Expanding flowfield Increased amplitude in
both ears
Table 1. Some sensorimotor contingencies associated with seeing and
hearing (taken from a draft version of O’Regan et al., 2005, available on-
line at http://nivea.psycho.univ-paris5.fr/CONS+COG/CC_OREGAN.htm).
While the work of O’Regan, Noë and Myin challenges the intuitive
and traditional concept of perceptual modalities and offers us a foun-
dation for a properly enactive approach to modality, the enactive
approach demands a somewhat clearer emphasis on the agent’s valued
actions. O’Regan & Noë’s approach is not challenged directly by the
evidence presented earlier on cross-talk between modalities, and the
work on TVSS is directly invoked by them in support of their theory.
Their account of modalities emphasises the interaction between the
agent and their environment, perceiving as exploration of the world
rather than a passive reception of information about the world. Given
such a view, there can be no primitives of perception simply given by
physiology — everything is in the interaction. What is more, that
interaction is always going to involve some aspect of goal-directed
activity on the part of the perceiving agent. It is such activity that
drives the exploration of the world in the first place, and gives value to
the interaction, the results of that exploration. It is this aspect of the
process is not quite unpacked in the detail it needs in O’Regan &
Noë’s work (nor indeed, in Noë, 2004), and so how it impacts on the
way we should consider perceptual modalities from an enactive point
of view needs a little more exegesis. In particular, while O’Regan &
Noë (2001a) and O’Regan et al. (2005) discuss the concept of modali-
ties characerised by sensorimotor contingencies, they do not fully
explore some of the more significant implications of a skills-based
account.
Remember, again, that within the enactive approach perception,
cognition and action, rather than being separable or clearly
ENACTING MODALITIES 85
distinguishable processes, are more like different aspects of the one
process of adaptive coping in which a goal-directed agent is continu-
ally involved. Modalities are not atomic in nature, but a product of a
dynamic process which involves an embodied agent (with goals and
sensitivities) and a world.
Within the traditional view, which holds modalities as basic ‘modes
of presentation’, a perception is simply ‘presented’ to us as either
visual, gustatory, tactile and so on. All other aspects of the perception
(recognition of the object, interpretation of the event) are deemed to
involve some form of further, often inferential, cognitive operation.
The enactive approach, however, rails against any such stage-like
description of perception, and in doing so transforms (and indeed
multiplies) the modes of experience available to us. Cognition is not
added to perception after the fact, because it is inherent in the process
of perception itself, it is part of what continually initiates, drives and
structures the act of perceiving.
An enactive approach to perception thus maintains a strong distinc-
tion between sensation and perception. Perception, wrapped up as it is
in cognition, action, sense-making, is an activity embedded within,
contextualised by, value-driven intentional action. Sensation is on
aspect of an embodied agent’s interaction with the world, an important
part certainly, but not one with any veto or absolute authority as the
character of experience.
The demand that perception be understood as structured by the
intentional actions of the agent as well as by the sensorimotor contin-
gencies that arise from embodied interaction with the world means
that the character of our perceptions is never, and can never be, simply
visual, or auditory, or tactile. In the abstract perceptual task of a
chess-player looking at a chessboard, for example, the player is not
only engaged visually with the world, but also engaged ‘chessily’.
Our perception of the world is imbued with goal-orientation and skil-
ful appraisal, such that the ‘mode of presentation’ will be as much a
matter of the skills we are deploying at the time as it is a matter of the
dynamics of sensory stimulation involved. Describing the perception
of the board as visual without describing it as chessy would be as
incomplete a description of the perception as describing the board
without making mention of its visual characteristics.
All perception, then, is inherently multi-modal. Our more abstract
skills cannot be deployed without engaging the embodied skills of
sensorimotor activity, but those sensorimotor skills are not encapsu-
lated either, they are engaged (their use structured) in the context of
goal-directed action. And of course, any normal perception will
86 M. McGANN
involve a plethora of such skills — in looking at a chess board I will
also be guiding visuomotor behaviours, engaged in a social interac-
tion and be prepared to use a whole manner of other skills depending
on the details of my context. The different ‘modes’ of perception,
then, are not sharply separated, but develop and operate together, to
the point that they can have quite deep dependencies. Many of us with
poor vision will have had the experience of having to put on our
glasses in order to hear more clearly, the interaction between taste and
smell is celebrated, in a game of soccer being able see the opening and
available run is not something that comes without being a skilled
player, but once you’re good enough, you ‘just see it’.3 As a final
example, I offer this suggestion as a phenomenological exercise for
the reader. Close your eyes and spin yourself around (either while
standing or, more safely, in a swivel chair) until you feel that you are
becoming quite dizzy. Stop completely, then open your eyes and fixate
an object as best you can. Your head movements and retinal stimula-
tion will be almost identical to a situation in which your continuing
maintenance of and sense of your balance was not so disturbed, but
what is your perception like? If you are anything like me, you will find
evidence that normal ‘visual’ perception has a surprising dependence
on vestibular activity.
Sensory substitution research, where sensory dynamics are trans-
formed by augmenting or compensating technologies, also supports
this view of modality. Adaptation to the technology in question only
occurs when the person engages in active, goal-directed behaviour.
This is the case for TVSS (Bach-y-Rita, 1972), and also for various
vision-to-audition devices that have been developed, such as Meijer’s
(1992) The Voice, which transforms pixel information from a camera
into a collection of audible frequencies (see also Auvray et al., 2007,
for an analysis of objection recognition and localisation with the
device).
Adaptation to the technologies also appears to be activity-specific.
For example, participants using the Voice, were more likely to claim
that their experience more resembled vision for object localisation
tasks, but more resembled hearing for object recognition tasks
(Auvray et al., 2007). Auvray & Myin (2009) argue that such sensory
substitution research (they prefer the phrase ‘perceptual
augmentation’) implies that perceptual modalities are a not rigid set of
ENACTING MODALITIES 87
[3] I am not a skilled soccer player – my position on the team could best be described as ‘hand-icap’, in that I’m placed on the team with better players when we can’t make even num-bers. I’m sure that all of us have comparable experiences where it is clear that our friendscan ‘just see’ or ‘just hear’ something – it pops out to them – that is utterly obscure to us.
distinct categories but are more like a space of possibilities. Some
areas of this space are more populated than others (the ‘traditional’
modalities), but intermediate forms of perception are also possible.
Some participants in Auvray et al‘s work with the Voice suggested
that rather than experiencing something visual or auditory, their expe-
riences when localising objects with the device was closer to a ‘new
sense’. In the case of TVSS, there is a similar debate over whether the
new experience should be considered ‘visual’ per se (see for example,
Prinz’s, 2006, criticisms of Noë, 2004, but also O’Regan & Noë’s
2001, p.958, p.1013 discussion of TVSS as ‘quasi-vision’) but this is
only a concern if we have good reason to believe that modalities must
be modular and discrete, which it appears they are not.
An enactive approach to modalities does not, therefore, make
strong or exclusive distinctions between forms of perception. Rather,
modalities are areas of stability within this space of possibilities,
stabilities that form on the basis of multiple interacting constraints —
the sensitivities of the individual agent, their goals, their expertise.
Stability does not imply fixation or rigidity, however, and even these
enacted stabilities will be in continual development, as our sensitivi-
ties, goals and expertise change. Our experiences are stable because
this development is very slow relative to our on-going actions.
For many, this may all sound a bit too idiosyncratic. Fine, we might
train or develop our visual or auditory skills in various ways, but there
remains something fundamental, something basic, which we all share
and which characterises some aspects of our perception no matter
what our goals, intentions and activities. We may see in different ways
in different activities, but it is precisely what is similar across the
experience of those different activities that allows us to identify some-
thing like vision. Red is red is red, after all, and whether I’m seeing red
because I’m checking which bishop my opponent has ready to move
against my king, or because I’m judging whether it is the same colour
as the car I’m thinking of buying, or whether I’m deciding if I like it,
my visual perception is still of the same red.
For an enactive account too, vision is precisely a set of aspects of
our experiences that are similar across different intentional contexts,
but without those different contexts there would be nothing to identify
as visual. Identifying a perceptual modality is not basic or fundamen-
tal, but is something done in the very same way as other forms of per-
ceptual judgement — with figure against ground. The enactive
approach simply makes this interdependency in perception more
explicit and drives home the importance of always identifying both
aspects of it.
88 M. McGANN
But, a critic might continue, having identified a continuity or con-
sistency across contexts that we can label as the ‘visual’ aspects of
experience, can we not then state what it is that underlies that consis-
tency or continuity? The answer is that we cannot, not in the categori-
cal terms that such a critic would find satisfying. Perceptual
modalities are modes of interaction, fuzzy sets of activities structured
by skilful action, environmental affordance and biological sensitivity
and irreducible to any of those components. Visual perceptions can be
identified not by the organs or biology involved, but by the kinds of
actions that they make possible, the kinds of activities in which they
evoked. It is a characteristic not of the biology of the agent, but of the
interaction in which the agent is engaged. That means that any set of
skills developed to a similar extent, be they concerned with vision,
hearing, driving, chess or social interaction, with have equal claim to
being a modality of perception.
This emergent nature of modalities, though, is not necessarily a
problem, and does not mean that we cannot effectively identify, com-
municate and investigate them. We human beings do share many
forms of sensitivity and many basic motivations. What is more, our
childhoods are long apprenticeships in which the development of our
perceptual skills are shaped and channelled so that significant varia-
tions or idiosyncrasies in experience are unlikely and unusual. Our
social skills play a part in the deployment of our visual, haptic and
olfactory ones, and the community in which we exist hold norms of
activity and norms of description which constrain our experiences. A
delightful and useful exploration of the development and disciplining
of experiences which acknowledges the interplay between the object,
the acting perceiver and the community of perception is Barry Smith’s
(2007) discussion of wine-tasting. Smith examines how our experi-
ences are guided and encouraged by a community which both
demands and validates certain kinds of interactions with the objects of
that community’s intentions. Something might be objectively the case
about the object, but not discriminated without the community’s guid-
ance and endorsement, a case where the modalities of taste and smell
are transformed by the development of a new ‘wine’ modality sup-
ported by an interacting society of experts.
While shared biological sensitivities enable this sharing of interac-
tions, they do not determine it. The story of how modalities are formed
and stabilised over developmental time is one which involves many
more players than biological maturation. Examining this question in
the kind of depth it deserves is unfortunately beyond the scope of this
paper. Nevertheless, the fundamentally social nature of that
ENACTING MODALITIES 89
development is worth a brief comment. While our bodies provide very
basic modes of activity from birth, the most potent of these is social
activity, a form of interaction which will become the primary means
by which many of our perceptual capacities will be scaffolded, cali-
brated and put to use. Social interaction is a complex coordination of
activity between two agents (De Jaegher & Di Paolo, 2007), involving
embodied interaction as well as the mastery of a host of ‘social’ con-
tingencies. These contingencies are not only sensorimotor, but pri-
marily emotional or affective in character. Due to the pervasive nature
of social skill and social interaction, its foundational role in the forma-
tion of our minds and its primacy in our interacting with and learning
about the world, it deserves at least as much recognition as a modality
as any of those with an easily identifiable peripheral organ (see
McGann & De Jaegher, 2009, for the first steps in an enactive analysis
of social skill and social perception).
Such a claim points to a final potential criticism of the enactive
approach to modalities: the idea that despite all of these tortuous
accounts of perception, the sensory organ really is some kind of final
arbiter on the subject. Other ways of skilfully interacting with the
world may be perceptual modalities, but certain modalities (five that
we could name immediately) are special, because they have sensory
organs dedicated and unique to them, while other skills are parasitic
on those basic, genuinely sensory modes of exploration of the world.
But this is just the same organ-centric criticism addressed above and is
subject to the same response — all of that dedicated sensation gets us
nowhere without reference to the appetite for exploration and action
that has been emphasised above. Sensory organs may be vital, but no
perception depends on, nor can be explained by, a specific organ
alone. Yes, vision is obviously dependent on the eyes, but it is also
non-obviously dependent in the normal case on vestibular function,
proprioceptive function (eyeball movements, for example), audition
and who knows what else? We should certainly take care to ensure
that the obvious facts are accounted for, but there is no call to restrict
our accounts to the obvious facts. Furthermore, given the active nature
of perception, without an account of the intention, the skilled activity,
driving the interaction between sensory organ and environment, we
will never find meaning in the patter of stimulation at the sensory
surface.
If, following the same organ-centric logic, we deny that the case of
a skilled TVSS user is a case of vision, however attenuated (none of
the vision-specific nerves have been enervated after all), we would
still be very hard-pressed to defend the claim that the experience is
90 M. McGANN
one of touch. We can identify perceptual modalities by disciplining
our discourse and calibrating our experiences with one another over
developmental time and through social practice, but we will not be
able to draw sharp and conclusive boundaries between them.
Implications of Enactive Modalities
No theory of <Insert your favourite modality>
One of the most dramatic implications of this approach to modalities
is that while it may be possible to produce a general theory of percep-
tion (as one aspect of a general theory of mind), we will not ultimately
be able to produce a general theory of any particular perceptual
modality. There is no generic or pure vision, no generic or pure hear-
ing, but specific and task-dependent interactions which involve visual
and auditory modalities.
Most research in these various domains will actually be unaffected
by this, as applied questions dominate. Where research questions and
methods are developed within a particular context they will produce
perfectly useful conclusions and generalisations within that domain.
But if our ambitions are to produce broad theories of an entire way of
being conscious — a theory of seeing, or of hearing — then we are
bound to be disappointed. The context-dependent nature of percep-
tion is not an interesting quirk of the perceptual system, but a funda-
mental characteristic of its operation. If we have learned nothing else
in the past century and a half of empirical research into the mind and
consciousness, it is that context is not contextual.
Far from restricting the ways in which we study perception, though,
this description of modality opens up a wide array of new perceptual
questions and domains of application.
Extending our understanding of perception and enactive
cognition
If the modalities of perception are a matter of the skills through which
we interact with the world, then the range of modalities which human
beings experience is not just greater than the traditional ‘five’, but is
in fact innumerable. Of course, we have always known that expertise
transforms a person’s perceptions in appropriate contexts, but for the
main the research within Cognitive Psychology has treated that phe-
nomenon as a post-perceptual process of inferential pattern-matching.
Experts have tens of thousands of situations represented in memory in
some way, and those patterns can short-circuit complex computational
demands to provide immediate guidance for behavioural output.
ENACTING MODALITIES 91
Except, if the enactive approach is correct then no such pat-
tern-matching goes on. Rather, as Hubert Dreyfus (2002) has argued,
drawing on Merleau-Ponty’s (1962) phenomenology of perception
and Freeman’s neurodynamical work, experts exhibit transformed
interactions in their chosen fields. Rather than matching patterns or
considering outcomes their experience of the situation is structured
within their expertise, such that rather than the expert responding to an
interpretation of the situation, the situation directly evokes or invites a
particular response, one that is structured from its foundations by that
person’s expertise.
The distinction between ‘lower’ and ‘higher’ cognition is therefore
a misleading one, and we should do away with it. Perception and cog-
nition are not sequential links in a chain but are fused, inseparable and
complementary aspects of the process of adaptive coping by an agent
in its environment. Within the enactive literature, this process of adap-
tive coping is referred to as ‘sense making’ the production of meaning
in interaction (Varela, 1991; Weber & Varela, 2002; Thompson &
Stapleton, 2008). Cognition is not to be understood as a set of pro-
cesses which occur in the brain over increasingly more complicated
representations but is the on-going activity of an entire organism
carrying out its goals within an environment that is, like a fulcrum,
simultaneously constraining and enabling that activity. Any account
of cognition that hopes to do the phenomenon justice must thus
encompass the entire system of the biology of the agent (their embodi-
ment), their goals, expertise and the environment. The consciousness
of this system will not be easily partialled into simple or universal
components, neither perceptual modality nor cognitive module. Any
aspect of an agent’s awareness during a particular action will have to
be described and interpreted in light of the rest of the system during
that same activity.
Conclusion
Once we make a clear and consistent distinction between sensation
and perception, the nature and structure of our awareness of the world
around us becomes dramatically affected. Psychological and
neuroscientific research have made the concept of a set of dedicated,
modular systems for the processing of sensory-specific information
problematic. What seems at first an issue in need of innocuous refine-
ment in fact reveals some significant fault lines in the traditional view,
and opens a series of questions which I suggest are best answered by
recourse to an enactive approach to understanding the mind. Fully
grasping the implications of an enactive account of modalities, we can
92 M. McGANN
see that the enactive approach need not, indeed cannot, be restricted to
the traditionally ‘lower’ forms of cognition. Nor is consciousness ever
a bare or simple phenomenon. Our goal-directed actions and skilful
cognition infuse and form consciousness from its very foundations
and our understanding of the processes of skilful, on-line embodied
coping may be used to transform and guide our investigations into
traditionally abstract and ‘higher’ cognition.4
References
Auvray, M., Hanneton, S. & O’Regan, J.K. (2007), ‘Learning to perceive with avisuo-auditory substitution system: Localization and object recognition withThe Voice’, Perception, 36, pp. 416–30.
Auvray, M. & Myin, E. (2009), ‘Perception with compensatory devices: From sen-sory substitution to sensorimotor extension’, Cognitive Science, 33, pp. 1036–58.
Bach-y-Rita, P. (1972), Brain Mechanisms In Sensory Substitution (London: Aca-demic Press).
Bach-y-rita, P. (1983), ‘Tactile vision substitution: Past and future’, InternationalJournal of Neuroscience, 19 (1), pp. 29–36.
Bach-y-Rita, P. (1984), ‘The relationship between motor processes and cognitionin tactile vision substitution’, In W. Prinz & A. Sanders (Eds.), Cognition andMotor Processes, pp. 149–50 (Berlin: Springer-Verlag).
De Jaegher, H. & Di Paolo, E. (2007), ‘Participatory sense-making: An enactiveapproach to social cognition’, Phenomenology and the Cognitive Sciences, 6,pp. 485–507.
Di Paolo, E. (2005), ‘Autopoiesis, adaptivity, teleology, agency’, Phenomenologyand the Cognitive Sciences, 4 (4),pp. 429–52.
Di Paolo, E. (2009), ‘Extended life’, Topoi, 28, pp. 9–21.Di Paolo, E.A., Rohde, M, & DeJaegher, H. (in press), ‘Horizons for the enactive
mind: values, social interaction and play’, In J. Stewart, O. Gapenne, & E. DiPaolo (Eds.), Enaction: Towards a New Paradigm of Cognitive Science (Cam-bridge, MA: MIT Press).
Dreyfus, H.L. (2002), ‘Intelligence without representation: Merleau-Ponty’s cri-tique of mental representation. The relevance of phenomenology to scientificexplanation’, Phenomenology and the Cognitive Sciences, 1 (4), pp. 367–83.
Freeman, W.J. (1991), ‘The physiology of perception’, Scientific American, 264,pp. 78–85.
Freeman, W.J. (2000), How Brains Make up Their Minds (London: Phoenix).Freeman, W.J. & Schneider, W. (1982), ‘Changes in spatial patterns of rabbit olfac-
tory EEG with condiitoning to odors’, Psychophysiology, 19 (1), pp. 44–56.Freeman, W.J. & Barrie, J. (1994), ‘Chaotic oscillations and the genesis of mean-
ing in cerebral cortex’, In G. Buzs’aki, R. Llin’as, W. Singer, A. Berthoz, & Y.Christen (Eds.), Temporal Coding in the Brain (Berlin: Springer-Verlag).
Hurley, S.L. & Noë, A. (2003), ‘Neural plasticity and consciousness’, Biology andPhilosophy, 18 (1), pp. 131–68.
ENACTING MODALITIES 93
[4] The author would like to thank three anonymous referees who made commentson two ear-lier drafts of this paper, and thanks to whose suggestions the current paper is muchimproved. I would also like to thank the members of the E- Intentionality seminar group atthe University of Sussex, who have discussed the ideas in this paper with the author onmore than one occasion.
Hurley, S.L. (1998), Consciousness in Action (Cambridge, MA: Harvard Univer-sity Press).
Kandel, E.R., Schwartz, J.H. & Jessell, T.M. (1995), Essentials of Neural Scienceand Behavior (London: McGraw-Hill).
McGann, M., & De Jaegher, H. (2009), ‘Self–other contingencies: Enacting socialperception’, Phenomenology and the Cognitive Sciences, 8 (4), pp. 417–37.
McGurk, H, & McDonald, J. (1976), ‘Hearing lips and seeing voices’, Nature,264, pp. 746–48.
Meijer, P. (1992), ‘An experimental system for auditory image representations’,IEEE Transactions on Biomedical Engineering, 39, pp. 112–21.
Merleau-Ponty, M. (1962), The Phenomenology of Perception (London:Routledge & Kegan Paul).
Myin, E, & O’Regan, J.K. (2002), ‘Perceptual consciousness, access to modalityand skill theories: A way to naturalize phenomenology?’, Journal of Conscious-ness Studies, 9 (1), pp. 27–45.
Newton, N. (1996), Foundations of Understanding (Amsterdam: John BenjaminsPublishing Co.).
Noë, A. (2004), Action In Perception (Cambridge, MA: MIT Press).O’Regan, J.K., Myin, E. & Noë, A. (2005), ‘Sensory consciousness explained
(better) in terms of “corporality” and “alerting capacity”’, Phenomenology andthe Cognitive Sciences, 4 (4), pp. 369–87.
O’Regan, J.K., & Noë, A. (2001a), ‘A sensorimotor account of visual conscious-ness’, Behavioral and Brain Sciences, 11 (5), pp. 939–73.
O’Regan, J.K. & Noë, A. (2001b), ‘What is it like to see: A sensorimotor theory ofperceptual experience’, Synthese, 129, pp. 79–103.
Rosenblum, L., Schmuckler, M. & Johnson, J. (1997), ‘The McGurk effect ininfants’, Perception and Psychophysics, 59, pp. 347–57.
Shams, L., Kamitani, Y. & Shimojo, S. (2000), ‘What you see is what you hear’,Nature, 408, p. 788.
Shams, L., Kamitani, Y. & Shimojo, S. (2001), ‘Sound alters visual evoked poten-tials in humans’, Journal of Vision, 12, pp. 3849–52.
Shimojo, S. & Shams, L. (2001), ‘Sensory modalities are not separate modalities:plasticity and interactions’, Current Opinion in Neurobiology, 11, pp. 505–509.
Thompson, E. (2007), Mind in Life: Biology, Phenomenology and the Sciences ofMind (Harvard University Press).
Thompson, E. & Stapleton, M. (2009), ‘Making sense of sense-making: Reflec-tions on enactive and extended mind theories’, Topoi, 28, pp. 23–30.
Varela, F.J. (1979), Principles of Biological Autonomy (Norwalk, CT: Appleton &Lange).
Varela, F.J. (1991), ‘Organism: A meshwork of selfless selves’, In A. Tauber (Ed.),Organism: The Origins of Self (Dordrecht: Kluwer).
Varela, F.J. (1997), ‘Patterns of life: Intertwining identity and cognition’, Brainand Cognition, 34 (1), pp. 72–87.
Varela, F. J., Thompson, E. & Rosch, E. (1991), The Embodied Mind (Cambridge,MA: MIT Press).
Violentyev, A., Shimojo, S. & Shams, L. (2005), ‘Touch-induced visual illusion’,Neuroreport, 16, pp. 1107–11.
Weber, A. & Varela, F.J. (2002), ‘Life after Kant: Natural purposes and theautopoietic foundations of biological individuality’, Phenomenology and theCognitive Sciences, 1 (2), pp. 97–125.
Paper received May 2008, revised June 2009
94 M. McGANN